Magnetic resonance image processing apparatus and method to which improvement in slice resolution is applied

Abstract

According to an embodiment of the present invention, there is provided a magnetic resonance image processing method, the magnetic resonance image processing method being performed by a magnetic resonance image processing apparatus, the magnetic resonance image processing method including: inputting an input image having a slice resolution higher than 0 and lower than 1 to an artificial neural network model; and outputting an output image having a slice resolution of 1 from the artificial neural network model.

1 .- 12 . (canceled) 
     
     
         13 . A magnetic resonance image processing method, the magnetic resonance image processing method being performed by a magnetic resonance image processing apparatus, the magnetic resonance image processing method comprising:
 inputting an input image obtained by accelerated imaging, having a slice resolution higher than 0% and lower than 100% to an artificial neural network model; and   outputting an output image having a slice resolution of 100% from the artificial neural network model,   wherein the slice resolution is defined as a relative size of k-space data range measured in the slice stack direction (Kz) in k-space, corresponding to the input image obtained through accelerated MRI scanning.   
     
     
         14 . The magnetic resonance image processing method of  claim 13 , wherein the artificial neural network model has been trained with training input data including image data having a slice resolution higher than 0% and lower than 100% and training label data including image data having a slice resolution of 100%. 
     
     
         15 . The magnetic resonance image processing method of  claim 14 , wherein:
 the training input data includes image data having a slice resolution higher than 0% and lower than 100% that is measured based on a setting of a slice resolution of the input image to 100%; and   the training label data includes image data having a same slice resolution as the input image.   
     
     
         16 . The magnetic resonance image processing method of  claim 14 , wherein:
 the training input data includes a training image and a slice image adjacent to the training image; and   a combination of the training image and the slice image adjacent to the training image is input to the artificial neural network model, and the artificial neural network model is trained on the combination.   
     
     
         17 . The magnetic resonance image processing method of  claim 13 , wherein:
 the input image includes a target image for which a slice resolution is to be increased and a slice image adjacent to the target image; and   a combination of the target image and the slice image adjacent to the target image is input to the artificial neural network model.   
     
     
         18 . The magnetic resonance image processing method of  claim 13 , wherein the input image includes at least one of a k-space image and a DICOM image. 
     
     
         19 . A magnetic resonance image processing apparatus, which is configured to input an input image obtained by accelerated imaging, having a slice resolution higher than 0% and lower than 100% to an artificial neural network model, and obtain an output image having a slice resolution of 100% from the artificial neural network model, and
 wherein the slice resolution is defined as a relative size of k-space data range measured in the slice stack direction (Kz) in k-space, corresponding to the input image obtained through accelerated MRI scanning.   
     
     
         20 . The magnetic resonance image processing apparatus of  claim 19 , wherein the artificial neural network model has been trained with training input data including image data having a slice resolution higher than 0% and lower than 100% and training label data including image data having a slice resolution of 100%. 
     
     
         21 . The magnetic resonance image processing apparatus of  claim 20 , wherein:
 the training input data includes image data having a slice resolution higher than 0% and lower than 100% that is measured based on a setting of a slice resolution of the input image to 100%; and   the training label data includes image data having a same slice resolution as the input image.   
     
     
         22 . The magnetic resonance image processing apparatus of  claim 20 , wherein:
 the training input data includes a training image and a slice image adjacent to the training image; and a combination of the training image and the slice image adjacent to the training image is input to the artificial neural network model, and the artificial neural network model is trained on the combination.   
     
     
         23 . The magnetic resonance image processing apparatus of  claim 19 , wherein:
 the input image includes a target image for which a slice resolution is to be increased and a slice image adjacent to the target image; and   a combination of the target image and the slice image adjacent to the target image is input to the artificial neural network model.   
     
     
         24 . The magnetic resonance image processing apparatus of  claim 19 , wherein the input image includes at least one of a k-space image and a DICOM image.